• <Plus d'infos sur intel.fr

Uncovering Secrets of the Universe with HPC

Veuillez nous excuser, ce PDF peut uniquement être téléchargé

French organization builds one of the world’s most powerful supercomputers using the Intel® Xeon® processor E5-2680

Grand Equipement National pour le Calcul Intensif (GENCI) is a government-owned organization in charge of coordinating high-performance computing (HPC) for France’s research institutes. It was created in 2007 to ensure France is at the international forefront of scientific research. GENCI is charged by the Ministry for Higher Education and Research with implementing and ensuring the coordination of the French national HPC centres by providing funding, assuming ownership, and promoting the use of HPC in fundamental and industrial research. GENCI is also the French representative of the Partnership of Advanced Computing in Europe (PRACE) and, as such, also has responsibility for driving HPC across Europe. With these objectives in mind, GENCI asked system integrator Bull to develop an HPC computing cluster based among others on the Intel® Xeon® processor E5-2680. The cluster consists of 92,000 cores installed and operated at the Très Grand Centre de calcul du CEA (TGCC) close to Paris.

Challenges
• Platform for discovery. GENCI wanted to build a powerful HPC platform to help scientists across Europe address challenging problems and advance scientific research
• Ambitious reach. Much of this scientific research requires enormous processing power – for example, to understand the evolution of the universe by simulating the distribution of dark matter throughout the full universe from the moment of the big ban

Solutions
• New technology. GENCI explored the benefits of the Intel Xeon processor E5-2680, which includes features designed to increase memory bandwidth and help double floating point performance compared to previous generation Intel Xeon processors
• Benchmarking software. It benchmarked the Intel Xeon processor E5-2680 using up to 12 real-life applications from various scientific domains. On some of them, like Monte Carlo computational algorithms, it achieved a 40 percent efficiency improvement and three-time peak performance improvement compared to previous generation Intel Xeon processor.